[m-dev.] [pma at miscrit.be: pfi server stats ...]

[Peter Ross]

Here are the first set of stats comparing the performance of the PFI server
under the different grades.

The important thing to note is that the hlc.gc grade is at most 10%
slower then the none.gc grade.
I assume that this slowdown is because twice as many garbage collections
occur, if we can get the GCs down to the same levels then the hlc.gc
grade should be significantly faster.

Paul and I are looking into that at the moment.

Pete

----- Forwarded message from Paul Massey <pma at miscrit.be> -----

From: pma at miscrit.be (Paul Massey)
To: petdr at mercury.miscrit.be
Subject: pfi server stats ...
Date: Fri, 28 Jul 2000 16:51:48 +0200


Taken from the server logfiles (each incomming request is timed) and the
results logged. This is the summary of those times.

    message name     cnt   tot time  TotGcs   AVtime (AvGcs)

pfiserver.none.gc.log

  createprocess:      601:  130390:     255 -  216   (0.42)
       propfind:      600:   24110:      51 -   40   (0.09)
      proppatch:      600:   87054:     194 -  145   (0.32)

pfiserver.hlc.gc.log

  createprocess:      591:  142127:     534 -  240 (0.90)
       propfind:      590:   24380:     102 -   41 (0.17)
      proppatch:      590:   90409:     379 -  153 (0.64)

Updates the the server will split these figures ...

Paul.


----- End forwarded message -----
--------------------------------------------------------------------------
mercury-developers mailing list
Post messages to:       mercury-developers at cs.mu.oz.au
Administrative Queries: owner-mercury-developers at cs.mu.oz.au
Subscriptions:          mercury-developers-request at cs.mu.oz.au
--------------------------------------------------------------------------